Emerging Markets Finance and Trade

ISSN: 1540-496X (Print) 1558-0938 (Online) Journal homepage: https://www.tandfonline.com/loi/mree20

Does Industrial Policy Play an Important Role in

Enterprise Innovation?

Feipeng Feng

To cite this article: Feipeng Feng (2019): Does Industrial Policy Play an Important
Role in Enterprise Innovation?, Emerging Markets Finance and Trade, DOI:
10.1080/1540496X.2019.1649654

To link to this article: https://doi.org/10.1080/1540496X.2019.1649654

Published online: 11 Sep 2019.

Submit your article to this journal

Article views: 9

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=mree20
Emerging Markets Finance & Trade, 1–23, 2019
Copyright © Taylor & Francis Group, LLC
ISSN: 1540-496X print/1558-0938 online
DOI: https://doi.org/10.1080/1540496X.2019.1649654

Does Industrial Policy Play an Important Role in

Enterprise Innovation?
Feipeng Feng
Business School of Guilin Tourism University, Guilin, Guangxi, China

ABSTRACT: Using a the sample of A-share listed companies in Shanghai and Shenzhen during the
twelfth five-year Plan period (2011–2015), this paper studies the impact of industrial policy and labor
allocation on the innovation efficiency of enterprises to test the correlation among government policy,
labor allocation and innovation efficiency. The empirical results show that industrial policy significantly
improves the innovation efficiency of enterprises; at enterprises with low labor allocation, the innovation
efficiency of enterprises is higher with an industrial policy. This study provides a new perspective and
empirical evidence that government decision-makers can use in dealing with the relationship between
government policies, labor market factors and enterprise innovation.
KEY WORDS: enterprise innovation, industrial policy incentives, labor cost, substitution effect
JEL: F229

Technological innovation is important for macroeconomic growth and for enterprises to create a long-
term competitive advantage. Since the global economic crisis in 2008, the world economy has
experienced great adjustment and reconstruction. In China, in 2016, during the thirteenth five-year
Plan (FYP; 2016–2020), economic development was supported by implementation of an innovative
development strategy. Few people pay for innovative knowledge after it is made public, which leads
to the failure of the knowledge market (Peneder 2008) in which innovators earn less than social
benefits.1 Because of the failure of the knowledge market and high risk (Acharya and Xu 2017; Kerr
and Nanda 2015), technological innovation requires well-functioning financial market to reduce
innovation costs. The existing literature offers many research results on the impact of government
policy incentives on micro-enterprise innovation behavior, mainly in three ways. First, papers explore
the impact of government support on R&D innovation activities (e.g., Li and Zheng 2016). Second,
case studies examine R&D behavior and its consequences caused by government support for
enterprise innovation (e.g., Ham and Mowery 1998; Vorley and Nelles 2010). Third, scholars
examine the impact of government support on the economic consequences of R&D innovation
(e.g., Aghion et al. 2012; Mukherjee, Singh, and Aldokas 2017; Song and Wang 2013). The main
motivation for enterprises to develop new technologies is to gain a competitive advantage by
reducing production costs (Rosenberg 1969; Spence 1984). However, the literature on the role of
government industrial policy and labor allocation as a measure of government resource allocation and
market-oriented factor allocation in influencing innovation efficiency of micro-enterprises is insuffi-
cient, and little empirical evidence is used to test this view and offer conclusions that corroborate
earlier efforts.
This paper investigates the impact of industrial policy and labor allocation on the innovation
efficiency of enterprises, Using a sample of A-share listed companies in Shanghai and Shenzhen
during the twelfth FYP period (2011–2015), we test whether a correlation exists among government
policy, labor allocation, and innovation efficiency. Our empirical results show that industrial policy
significantly improves the innovation efficiency of enterprises; at enterprises with low labor

Address correspondence to Feng Feipeng, Guilin Tourism University, 26 Liangfeng Road, Yanshan District,
Guilin, Guangxi 541006, China. E-mail: [email protected]
2 F. FENG

allocation, the innovation efficiency of enterprises is higher with the support of industrial policy; and
the mechanism regulating the role of labor allocation in industrial policy and enterprise innovation is
that industrial policy is conducive to stimulating enterprises to increase R&D investment. Enterprises
with a smaller labor allocation tend to have higher labor costs, and when labor costs are high, they
stimulate enterprises to increase R&D investment, thus showing the substitution effect on labor
investment of R&D investment. Because of market pressure from higher labor costs, the interaction
between industrial policy incentives and R&D investment greatly promotes the innovation efficiency
of enterprises.
We make the following contributions. First, our research focuses on the efficiency of innovation
(measured by the amount of applications for invention patents), which is beneficial the government’s
implementation of innovation-led economic policies. Second, our research perspective treats indus-
trial subsidies and tax incentives as incentives for industrial financial means, and to allocate labor
force (employee gross). The number divided by the scale of enterprise assets is mainly considered
from the market allocation power, aiming at exploring the different effects of government power and
market power on enterprise innovation efficiency. Third, based on the potential impact of labor
scarcity on enterprise innovation (Mao and Wang 2018), we investigate the innovation behavior of
enterprises and broaden the research to encompass macroeconomic policies and micro-enterprises.
Following this introduction, the paper is organized as follows. Section 1 reviews the literature and
outlines our research hypothesis. Section 2 describes our samples and data. Section 3 gives the
definition of the variables and descriptive statistics of main variables. Section 4 is our empirical
analysis, through regression analysis to examine the significance of statistical coefficients of variables
and confirm our hypothesis. Section 5 is the main conclusions and policy implications, describing the
significance of this study.

1. Literature Review and Research Hypothesis

An industrial policy, undertaken to support economic development as part of a national strategy,
directs the development of various industries through the use of methods, such as subsidies, credit,
prices, import and export control, foreign finance, examination and approval,2 directives, government
performance appraisals, autonomy,3 and information dissemination. In China, an industrial policy and
supporting measures are part of the government’s development strategy for economically catching up
with the advanced economies. The Chinese government use industrial policy to allocate resources
effectively to sectors with higher production efficiency, so as to promote long-term economic growth.
The government adjusts the industrial structure, organization and layout, promotes industrial upgrad-
ing and technological progress, and injects new impetus into economic development through the
industrial policy and the market.
Industrial policy can induce the capital flow of enterprises by influencing the resource availability
of micro-enterprises and the business environment of enterprises. Industrial policy guidance and
incentives (e.g., industrial subsidies and tax incentives) can play a role in promoting R&D investment
in the following areas. First, industrial policy can play the necessary policy compensation role for
R&D. In innovation, public policy to provide benefits and compensation at all levels of governance —
for example, targeted credit, R&D subsidies, and funding for technological upgrading and transfor-
mation of traditional industries. This helps to bridge the gap between private benefits and social
benefits caused by the partially (or totally) public goods character of innovative activities by
innovative enterprises, thus promoting innovative investment by enterprises.
Second, industrial policy guidance helps to alleviate the negative impact of information asymme-
try. Economic theory provides a good reason for industrial policy to stimulate innovation investment.
Specific industrial policy support measures, such as industrial subsidies and tax incentives, are
a policy means not only to guide enterprises to adjust the allocation of resources within and between
industries, but also to release an economic signal to the market about the different tendencies in
 DOES INDUSTRIAL POLICY PLAY AN IMPORTANT ROLE IN ENTERPRISE INNOVATION? 3

resource allocation among different industries, which may be conveyed by whether industrial sub-
sidies and tax incentives are given. This can effectively reduce the concerns among micro-market
participants about policy uncertainty.
Third, industrial policy potentially allocates resources differently, which benefits enterprises
supported by the policy and helps to reduce their operating costs. It is an indirect incentive for
enterprises to invest in innovation. Industrial policy offers financing convenience for enterprises
supported by it, indicating an economic signal by the government, and the degree of support for
different industries shows the orientation of national industrial development. In addition, industrial
policy can also reduce information asymmetry between banks and enterprises, better establish trust
and cooperation between banks and enterprises (Chen, Li, and Xin 2013), and affect the development
prospects and investment behavior of enterprises (Lu and Han 2013), which is conducive to better
development of those enterprises. The policy signal sent by industrial policy is the correction of
market information failure. By selectively supporting several key industries as leading industries, the
government transmits strategic information on industrial development to the market, reduces the
information cost for banks in searching for enterprises with demand for high-quality credit, guides
funds to enterprises with development potential, and provides financial support for technological
innovation. It also promotes the transformation of individual technological innovation enterprises
from “innovation numbness” to “joint innovation”,4 thus driving the technological innovation and
industrial upgrading of enterprises. Therefore, industrial policy can greatly reduce the cost of
information collection, the asymmetry of market information, and transaction costs; it can effectively
reduce uncertainty in economic activities before and after, and have a positive impact on enterprise
innovation investment. Under the guidance of industrial policy, the increase in innovation investment
helps reduce enterprise innovation risk and increase the potential for innovation output. Based on this
analysis, we propose our first hypothesis:
Hypothesis 1: Industrial policy improves innovation efficiency by enterprises.
In addition, market factors influence the effect of industrial policy in promoting enterprise
innovation. For example, labor scarcity is an important economic channel affecting enterprise
innovation. When the allocation of labor is low, it often means that the cost of labor is high, so
enterprises have a strong internal motivation to adopt labor-saving technology (Mao and Wang 2018).
For example, in the early nineteenth century, when labor costs were high, companies tried to invent
labor-saving technologies and reduce costs by reducing enterprise resource consumption, thus
providing incentives for R&D innovation (Spence 1984). During the period before the Civil War in
the United States, labor scarcity clearly differed between the slave states in the South and the free
states in the North. Slave labor in the Southern States had a much lower marginal cost than the “free
labor” in Northern States, so Northern states had a greater labor shortage. The gap in labor scarcity
was significantly widened by the lack of labor mobility across regions. Research shows that labor
scarcity between Southern and Northern states stimulated the role of finance in technological
innovation of enterprises to a greater extent, such as efficiency measured by patent activities,
especially in the North, where labor was relatively scarce.
In addition, the model built by Acemoglu (2010) links the labor force, enterprise financing, and
a balanced innovation level. In this model, the economy has a monopoly entrepreneur, innovating
new technologies to make machines that contain patented technology. End -product producers
combine labor and new technologies to maximize profits by optimizing the demand for labor and
machinery. Entrepreneurs pay the cost of manufacturing new technologies and gain a monopoly. The
innovation level and monopoly price of the machines are selected based on demand for them. To
some extent, the labor shortage stimulates demand for labor-saving technology.
Technology can complement labor (e.g., in education, computer-assisted instruction). As seen in
some noteworthy examples throughout history. For example, in the nineteenth century, technological
progress in the United States was characterized by labor scarcity— that is, technology that conserved
4 F. FENG

labor was more widely used as a driving force in innovation. The shortage of American workers was
a greater incentive for American entrepreneurs to replace labor with machines than for their British
counterparts (Habakkuk 1962). In comparing labor-saving and labor-complementary technologies,
Acemoglu (2010) believes that it is likely that technological progress in Britain and the United States
in the late eighteenth and nineteenth centuries were strongly based on saving labor. Mao and Wang
(2018) studied and found that labor-saving technology is an alternative to productive labor. Labor
shortages prompt enterprise managers to demand labor-saving technology and enhance financial
influence on innovation. This view is consistent with Hicks’s (1932) view that changes in the relative
price of production factors are incentives for invention and for an invention to conserve on
components that have become relatively expensive. Acemoglu (2002, 2010) theoretically demon-
strates the role of factor prices in guiding innovation and technological change. Bena and Simintzi
(2017) studied the 1999 US-Chinese bilateral agreements and found that cheap offshore workers
discouraged labor-saving innovation. Tuzel and Zhang (2018) found that investment tax incentives
encourage computer purchases and reduce routine tasks. The history of economic development
provides additional evidence, such as the sudden decline in the labor supply caused by flooding of
the Mississippi River in 1927 and the decline in cotton exports to Britain caused by the American
Civil War (Hanlon 2015; Hornbeck and Naidu 2014). Based on this discussion, we propose
our second hypothesis:
Hypothesis 2: Enterprises with low labor allocation have higher innovation efficiency with the support of
an industrial policy.
Next, we elaborate on the mechanism of labor force allocation in industrial policy and enterprise
innovation. First, the input of innovation activities is restricted by two important factors that affect the
innovation of enterprises: the attribute of all or part of public goods of innovation output,5 the non-
exclusive (positive) externalities of innovation income resulting from it that private gains are smaller
than social gains; and the defects of capital markets (information asymmetry, difficulty in identifying
the risk of innovation project, etc.) (Peneder 2008), that is, insufficient incentive for interest. And the
constraints caused by financing constraints, resulting in the failure of the knowledge market and
inadequate R&D investment. Industrial policy guidance (support and non-support) means that the
government sends out an economic signal of industrial intervention, which is not only a kind of
information disclosure, but also a kind of formal/informal (explicit/implicit) contract. The guidance
of industrial policy needs to be implemented by means of benefit-driven means (e.g., industrial
subsidies, tax preferences,), so as to send a precise and credible signal to the market, thereby reducing
information asymmetry and transaction costs, guiding investors’ expectations, and realizing external
incentives for enterprise managers, which is conducive to making up for the deficiencies of innova-
tion investment in interest incentives and investment. Enterprise managers are encouraged to increase
R&D investment in order to form a long-term competitive advantage (Lazzarini 2015).
Second, the scarcity of different resources is often reflected in different price or cost effects, thus
creating a certain market pressure. As mentioned above, the different allocation of the labor force is
often a market response to labor cost signals. Enterprises with a lower labor allocation often have
higher labor costs. For example, in recent years, when a shortage of first-line production workers
arises in the developed regions of China (Yangtze River Delta, Pearl River Delta, etc.), the labor cost
has risen sharply, which has prompted enterprise managers to adopt advanced intelligent technology
or industrial robots to replace workers’ labor. That is, the market pressure of labor cost urges
enterprises to adopt labor-saving technology. At the same time, when the price of labor factors is
high, the change in the relative price of a production factor is a stimulus to invention, and is for
a specific type of invention, in order to substitute for a component that has become more expensive
(Hicks 1932). This means that when the cost of labor is high, enterprises will be stimulated. The
industry increases R&D investment more, which shows the substitution effect of R&D investment on
labor investment. Under market pressure for higher labor costs, the interaction between industrial
 DOES INDUSTRIAL POLICY PLAY AN IMPORTANT ROLE IN ENTERPRISE INNOVATION? 5

policy incentives and R&D investment promotes improvement in innovation efficiency by enter-
prises. Therefore, based on this analysis, we propose our third hypothesis:
Hypothesis 3: Enterprises with lower labor allocation and higher labor costs but with higher R&D
investment have an improvement in innovation efficiency.

2. Research Samples and Data

2.1. Selection of the Sample Period
The reasons for choosing 2011–2015 as the sample period is that the previous FYP period
(2006 − 2010) included, the global financial crisis in 2008, which affected Chinese exports and led
to an economic decline. The Chinese government invested about RMB 4 trillion over the period
2008–2010 to maintain steady growth. This investment achieved only short-term results because the
funds were spent without any planning or project support, in an arbitrary fashion. Since 2010, China’s
economy has fallen sharply, even more than in 2008. Therefore, we selected 2011–2015 to better
study consistency in the orientation and implementation of industrial policies. In addition, the most
recent FYP covered 2011–2015, so it has the most up-to-date data.

2.2. Data Selection and Processing

To investigate the impact of industrial policies and financial means (industrial subsidies, tax
incentives, etc.) on R&D investment by enterprises, we collected data on all A-share companies
listed on the Shanghai and Shenzhen Stock Exchanges during the twelfth FYP period based on the
industry classification standards of listed companies by the China Securities Regulatory
Commission. The data come from China Stock Market and Accounting Research and Wind
databases.
To reduce the adverse effects of extreme values on statistical results, we winsorize the extreme
values in the 1% and 99% quantiles to deal with the continuous variables involved in the study.
Considering the robustness of the research results, according to the usual practice, we deleted
financial companies, ST and PT6 companies and companies with shareholder equity of 0 or less;
company data in the year of the initial public offering; and companies with incomplete or missing
financial data during the sample period. This yields annual sample observations of 4,361 listed
companies.

3. Definition and Descriptive Statistics of the Main Variables

In this section, we define and measure the main variables in this paper, so as to describe the specific
operational procedures in our empirical research. In addition, we list the basic descriptive statistics of
the important variables to be investigated in this paper so as to reveal the preliminary statistical
characteristics among variables.

3.1. Definitions of the Main Variables

This paper studies the effect of industrial policy and its financial means (e.g., subsidies, tax
incentives) on innovation efficiency. We rely on Zhou et al. (2015) and Li and Zheng (2016) in
constructing the following variables.
Innovation efficiency (invent) is measured as the annual number of applications for invention
patents, to indicate substantial technological progress.
Industrial policy variables (ip) are dummy variables. When the industry during the twelfth FYP
was supported by the government, the variable is set at 1; otherwise, the variable is set at 0.7
6 F. FENG

Industrial Subsidies (subsidy) are measured as government industrial subsidies/business income.

Our definition follows Li and Zheng (2016) and Zhou et al. (2015). Industrial subsidies are a common
means of fiscal support for industrial development by the government.
We follow Aghion et al. (2012) and Song and Wang (2013) in defining tax incentives as follows:
Tax incentives = [(statutory basic income tax rate - effective income tax rate) × pre-tax profits]
divided by business income.
In this paper, the statutory basic tax rate refers to the income tax rate for ordinary enterprises
(25%) under the Income Tax Law of January 1, 2008, which is lower than the basic tax-ratio,
therefore it is a preferential tax rate. Following Mao and Wang (2018), we define Labor as the
number of employees/total assets (in RMB 100 million) measure the allocation of labor in enter-
prises. We also consider industries and years as dummy variables. Table 1 lists the definition and
measurement of the variables.

Table 1. Definitions of the variables.

Variable Variable label Variable definition

Innovation efficiency Invent Substantial technological progress by the enterprise, measured as the
annual number of applications for invention patents.
Industrial policy Ip If the industry year is supported by the government during the twelfth FYP,
Ip is set at 1; otherwise, Ip is set at 0.
Industrial subsidy Subsidy Measured as government industry subsidies divided by enterprise operating
income.
Tax incentives tax-incentives Measured as [(statutory basic income tax rate – effective income tax rate of
enterprises) × pre-tax profit] divided by business income.
Labor allocation Labor Measured as the number of firm employees of the company divided by total
firm assets of the company (in RMB 100 million).
Leverage Leverage Measured as total liabilities divided by total assets to characterize the state
of the company’s financial status.
Share of State- state- share Measured as the number of state-owned shares divided by total shares, to
ownership characterize the degree of state control over corporate equity.
Market share market-share Measured as the ratio of the firm’s annual sales revenue to the total sales
revenue of the entire industry.
Company size Size Measured as the natural logarithm of the total asset value at the beginning
of the period.
Diversified operations diversification If the value of SR is more than 95%, then diversification = 0; otherwise 1, it
is diversified. American scholar Wrigley (1970) proposed the concept of
specialization ratio. Measured as: Specialization rate (SR) = sales of the
company largest business project divided by total sales of the company.
Ability to repay debts ability-to-repay Measured as (total fixed assets - cumulative depreciation - fixed assets
impairment provision) divided by total assets.
Equity concentration e-concentration Measured as the shareholding ratio of the largest shareholder divided by the
shareholding ratio of the second-largest shareholder
Company-control- control- Measured as the sum of the shareholding ratios of the second to fifth
competition competition shareholders
Proportion-of- independence Measured as the number of independent directors on the board of a listed
independent- company divided by total number of board directors of listed companies.
directors
Separation of power separation Measured as the difference between control and ownership, that is, the
actual controller holds the ratio of the voting right in company decision-
making - the actual controller has the ratio of shares in a company.
Years Year dummy variable
Industry Indu dummy variable
 DOES INDUSTRIAL POLICY PLAY AN IMPORTANT ROLE IN ENTERPRISE INNOVATION? 7

3.2. Descriptive Statistics of the Main Variables

Our sample period (2011–2015) has total observations of 4,361 listed companies. However, because
of incomplete data on some companies, our descriptive statistics (see Table 2) are missing data
(unbalanced panel).
The average value of invent is 10.44, and the range is 3,559, which shows that the individual
differences in enterprise innovation are very large. At the same time, ip of the full sample has an
average value of 0.643, which means that on average 64.3% of the study sample is supported by
industrial policies, which indicates that during the twelfth FYP period, industrial policy is mostly
under the control of the central government.
The mean and median of industrial subsidies (subsidy) are 1.399, and 0.974, respectively; and the
mean and median of tax preferences (tax-incentives), are 0.710, and 0.675, respectively. This shows
that the sample listed companies’ subsidies have a higher level than tax incentives. The average for
labor is 96.82, which shows that during the sample period, the labor force allocation of an enterprise
is with 96.82 employees per asset of RMB 100 million.

4. Empirical Analysis
In this section, we conduct an empirical analysis of our model results. First, based on the full sample,
we examine the industrial policy propensity and support mechanism (industrial subsidies, tax
incentives), and differences in the efficiency of enterprise innovation. Second, we explore whether
supported enterprises play a regulatory role in the relationship between industrial policy and
corporate efficiency. Third, we looked at the mechanism of the adjustment of labor allocation in
industrial policy and enterprise innovation efficiency.

4.1. Model Design

Models (1), (2), and (3) are designed to test whether industrial policy (ipi, j, t) guidance and its
specific financial means (industrial subsidies, tax incentives) make industrial policy support for
enterprise innovation more efficient.

lninventi;j;t ¼ a0 þ a1 ipi;j;t þ CVi;j;t þ ei;j;t (1)

Table 2. Descriptive statistics of the main variables (N = 4,361).

Variable Mean Sd p50 min max Range (= max.-min.)

invent 10.44 68.09 2 0 3,559 3,559

ip 0.643 0.668 1 0 1 1
subsidy 1.399 1.918 0.710 0 10.80 10.80
tax-incentives 0.974 1.271 0.675 0 5.789 5.789
labor 96.82 71.71 79.00 1.026 596.9 595.9
leverage 39.71 21.58 38.25 0.797 180.6 179.8
state- share 3.396 11.26 0 0 83.68 83.68
market-share 3.295 1.073 3.111 1.686 5.768 4.083
size 0.470 0.499 0 0 1 1
ability-to-repay 21.74 13.23 19.35 1.294 54.04 52.75
e-concentration 9.571 13.11 3.920 1.095 53.73 52.64
control-competition 18.41 11.18 16.85 2.130 40.24 38.11
independence 37.29 5.085 33.33 33.33 50 16.67
separation 4.977 7.307 0 0 21.60 21.60
8 F. FENG

lninventi;j;t ¼ b0 þ b1 ipi;j;t þ b2 subsidyi;j;t 1 þ b3 ipi;j;t  subsidyi;j;t 1 þ CVi;j;t þ ei;j;t (2)

lninventi;j;t ¼ c0 þ c1 ipi;j;t þ c2 tax  incentivesi;j;t 1 þ c3 ipi;j;t  tax  incentivesi;j;t 1 þ CVi;j;t

þ ei;j;t (3)

where i, j, and t represent the observed value of the jth enterprise in the ith industry in the year
t. For simplicity, we omit the subscripts in the variables in the tables, here and below. The dependent
variables in model (1), (2), (3) are lninventi, j, t, and the main independent variables are ipi, j, t, and
interaction terms are ipi,j,t × subsidyi,j,t _1 and ipi,j,t × tax-incentivesi,j,t _1. We expect a1, b3, and c3 to
be more than 0, that is, industrial policies help enterprises to have higher innovation efficiency.
Models 4 and 5 test H2 that is, at those enterprises with low labor allocation, industrial policy helps to
increase innovation efficiency of enterprises. In this paper, we divide the groups following Aghion,
Hleimous, and Kharroubi (2014), when the value of each characteristic index(labor) is higher than the
median value group, it is in the high value group H; otherwise, it is in the low value group L. First, we test
these models by grouping the regressions and observe the differences in coefficients of the core independent
variable coefficients ipi,j,t × subsidyi,j,t _1, ipi,j,t × tax-incentivesi,j,t _1 in different groups. Then we examine
the differences in the values of subsidyi,j,t _1 × labori,j,t _1, tax-incentivesi,j,t _1 × labori,j,t _1 in the sample
supported by industrial policy (ip = 1). The positive and negative signs of the coefficients are used to obtain
the different effects of labor allocation factors on the marginal effects of industrial policy on enterprise
innovation when the labor allocation increases from the low value group L to the high value group H.

lninventi;j;t ¼ a0 þ a1 subsidyi;j;t 1 þ a2 labori;j;t 1 þ a3 subsidyi;j;t 1  labori;j;t 1 þ CVi;j;t

þ ei;j;t (4)

lninventi;j;t ¼ a0 þ a1 tax  incentivesi;j;t 1 þ a2 labori;j;t 1 þ a3 tax  incentivesi;j;t 1

 labori;j;t 1 þ CVi;j;t þ ei;j;t (5)

To obtain more statistical details in the empirical results, we add quadratic equations with interaction
terms (subsidyi,j,t _1 × labor i,j,t _1)2 and (tax-incentivesi,j,t _1 × labori,j,t _1)2 in models (4) and (5) to test
whether labor allocation, as a market factor, has an inverted U-shaped relation to the marginal effect of
industrial policy on enterprise innovation.
To further reveal the mechanism of the role of labor allocation in the relationship between
industrial policy and enterprise innovation efficiency, i.e. to examine the channels through which
labor allocation plays a regulatory role in industrial policy and enterprise innovation risk, or to
confirm H3 is valid.

4.2. Empirical Test Resuts

The results of models are in Table 3–5 respectively.
Table 3 shows the relationship between China’s industrial policy and enterprise innovation. In
Table 3, the dependent variable is lninvent, the core independent variable are ip, ip×subsidy_1,
ip×tax-incentives_1 and the other control variables as defined in Table 1.
In Table 3 column (1), the regression coefficient of enterprise innovation efficiency (lninvent)
on industrial policy (ip) is significantly positive (0.163). That is, for every 1% increase in the
government’s share of supporting enterprises, the innovation efficiency of enterprises increases
by 16.3%. This shows that the innovation efficiency of enterprises supported by industrial
policy is higher than that of enterprises not supported by industrial policy. After being added
to the benefit incentive mechanism (industry subsidy and tax-incentives), the coefficients of
 DOES INDUSTRIAL POLICY PLAY AN IMPORTANT ROLE IN ENTERPRISE INNOVATION? 9

Table 3. Impact of industrial policy on patent output (based on the full sample).
lninvent
Full sample (1) (2) (3)

ip 0.163*** 0.0943*** 0.147***

(5.21) (2.64) (3.02)
subsidy _1 0.0358**
(2.38)
ip×subsidy _1 0.0492***
(2.96)
tax-incentives _1 0.0959***
(3.23)
ip×tax-incentives _1 0.119***
(7.20)
leverage -0.00613*** -0.00509*** -0.00375***
(-5.55) (-4.59) (-2.87)
market-share -0.278** -0.235* -0.213
(-2.10) (-1.77) (-1.26)
state- share -0.00291* -0.00304* -0.00282*
(-1.71) (-1.79) (-1.92)
size 0.389*** 0.389*** 0.339***
(13.57) (13.62) (9.72)
diversification -0.261*** -0.261*** -0.0484
(-6.75) (-6.80) (-1.26)
ability-to-repay -0.00417*** -0.00393*** -0.00252
(-2.79) (-2.64) (-1.54)
e-concentration -0.00481** -0.00485** 0.0000780
(-2.42) (-2.45) (0.04)
control-competition -0.000833 -0.000866 0.000302
(-0.36) (-0.37) (0.12)
independence -0.00196 -0.00383 -0.000614
(-0.53) (-1.03) (-0.18)
separation -0.00151 -0.000752 -0.0000620
(-0.57) (-0.29) (-0.02)
_cons 0.612*** 0.570*** 0.321*
(3.37) (3.15) (1.68)
Year &industry Controlled Controlled Controlled
Obs. 4361 4361 4361
R-squared 0.0651 0.0770 0.0840
Adj. R-squared 0.0601 0.0701 0.0802

t statistics are in parentheses. * p < 0.10, ** p < 0.05, *** p < 0.01.

ip×subsidy_1 and ip×tax-incentives_1 are significantly positive (0.0492 and 0.119, respectively)
in columns (2) and (3) of Table 3. That is, compared to enterprises that are not supported by
industrial policy, the Industrial Subsidy and income tax preferences of enterprises supported by
industrial policy increase by 1%, and the innovation efficiency of enterprises increases of
0.0492% and 0.119% respectively. This confirms H1.
Table 4 is a group regression based on different levels of labor allocation, which is used to test the
relationship between industrial policies and enterprise innovation in different groups of labor alloca-
tion. we divide the groups following Aghion, Hleimous, and Kharroubi (2014), when the value of
each characteristic index(labor) is higher than the median value group, it is in the high value group
10 F. FENG

Table 4. The impact of industrial policies on patent output (grouped regression results based
on the division of labor allocation)
lninvent

L-labor H-labor L-labor H-labor

(1) (2) (3) (4)

ip 0.0293 0.240*** 0.0797 0.336***

(1.27) (4.57) (1.23) (6.48)
subsidy _1 0.00195*** 0.0767***
(4.54) (2.90)
ip×subsidy _1 0.0390** 0.0356
(2.18) (1.19)
tax-incentives _1 0.0910** 0.392***
(2.37) (7.63)
ip×tax-incentives _1 0.0853*** -0.185***
(3.61) (-3.56)
leverage -0.00710*** -0.00489*** -0.00577*** 0.000461
(-3.65) (-3.26) (-2.86) (0.29)
market-share 0.0706 -0.816*** 0.106 -0.784***
(0.32) (-3.81) (0.48) (-3.71)
state- share -0.00292 -0.00512** -0.00301 -0.00440*
(-1.33) (-2.25) (-1.37) (-1.95)
size 0.332*** 0.533*** 0.326*** 0.499***
(6.38) (12.90) (6.26) (12.20)
diversification -0.0340 -0.246*** -0.0268 -0.241***
(-0.61) (-4.70) (-0.48) (-4.65)
ability-to-repay -0.00472** -0.00117 -0.00365 0.000961
(-2.11) (-0.53) (-1.62) (0.44)
e-concentration -0.000164 -0.00625** -0.000426 -0.00650**
(-0.06) (-2.20) (-0.16) (-2.32)
control-competition 0.00160 -0.00191 -0.0000447 -0.00410
(0.44) (-0.60) (-0.01) (-1.29)
independence -0.00555 -0.00463 -0.00459 -0.00271
(-1.08) (-0.93) (-0.90) (-0.55)
separation 0.00383 -0.00420 0.00433 -0.00654*
(0.76) (-1.18) (0.86) (-1.86)
_cons 0.632** 0.181 0.507* -0.136
(2.28) (0.72) (1.81) (-0.54)
Year &industry Controlled Controlled Controlled Controlled
Obs. 2180 2181 2180 2181
R-squared 0.0839 0.1116 0.0860 0.1304
Adj. R-squared 0.0812 0.1019 0.0821 0.1286

t statistics are in parentheses. * p < 0.10, ** p < 0.05, *** p < 0.01.

(H-labor); otherwise, it is in the low value group (L-labor). In Table 4, the dependent variable is
lninvent, the core independent variables are ip, ip×subsidy_1,and ip×tax-incentives_1 and the other
control variables are defined in Table 1.
In Table 4, we show the moderating effect on the relationship between industrial policy and
enterprise innovation efficiency after adding labor allocation factors. Columns (1) and (2) in Table 4
show that the coefficient of ip×subsidy_1 decreases from 0.0390 to 0.0356 when the labor allocation
increases from small to large, and changes from significant to insignificant. That is, when labor
Table 5. The impact of industrial policies on patent output (based on the regression results of the interaction between financial support and
labor allocation of enterprises supported by industrial policies).
lninvent

(1) (2) test of inverted U-shaped curve(3)

(a)L-labor (b)H-labor (a)L-labor (b)H-labor (a) (b)

subsidy _1 0.0337*** 0.139*** -0.0256

(5.41) (3.06) (-1.12)
tax-incentives_1 0.164*** 0.249*** 0.0394
(2.71) (3.70) (1.26)
labor_1 0.00837*** 0.000245 0.0117*** 0.0000917 0.000445 0.000999**
(3.84) (0.43) (5.01) (0.17) (1.05) (2.44)
subsidy _1×labor_1 0.00101*** -0.000227 0.00283***
(4.41) (-0.73) (7.21)
2
(subsidy _1×labor_1) -0.00000168***
(-7.23)
tax-incentives_1×labor_1 0.00239*** -0.000306 0.00208***
(4.99) (-0.75) (5.30)
(tax-incentives_×labor_1)2 -0.00000172***
(-5.43)
leverage -0.00594*** -0.00609*** -0.00334 -0.00157 -0.00611*** -0.00191
(-3.00) (-3.16) (-1.54) (-0.75) (-4.42) (-1.26)
market-share -0.531** -1.832*** -0.615** -1.832*** -1.042*** -1.108***
(-2.11) (-5.80) (-2.45) (-5.82) (-5.32) (-5.64)
state- share 0.000791 -0.00436* 0.00142 -0.00449* -0.00247 -0.00246
(0.25) (-1.66) (0.45) (-1.71) (-1.22) (-1.21)
size 0.474*** 0.638*** 0.462*** 0.595*** 0.521*** 0.468***
(9.18) (11.67) (8.87) (10.88) (14.04) (12.50)
diversification -0.205*** -0.152** -0.170** -0.142** -0.190*** -0.172***
(-3.08) (-2.28) (-2.53) (-2.14) (-4.05) (-3.65)
ability-to-repay -0.0137*** 0.00301 -0.0117*** 0.00546* -0.00673*** -0.00448**
(-5.02) (1.04) (-4.24) (1.87) (-3.40) (-2.23)
e-concentration -0.00151 -0.00528 -0.00234 -0.00503 -0.00305 -0.00422
DOES INDUSTRIAL POLICY PLAY AN IMPORTANT ROLE IN ENTERPRISE INNOVATION?

(-0.39) (-1.33) (-0.61) (-1.27) (-1.11) (-1.52)

(Continued )
11
 12
F. FENG

Table 5. The impact of industrial policies on patent output (based on the regression results of the interaction between financial support and
labor allocation of enterprises supported by industrial policies). (Continued)
lninvent

(1) (2) test of inverted U-shaped curve(3)

(a)L-labor (b)H-labor (a)L-labor (b)H-labor (a) (b)

control-competition -0.000123 0.00116 -0.00251 -0.000447 0.000220 -0.00219

(-0.03) (0.29) (-0.62) (-0.11) (0.08) (-0.76)
independence -0.00167 -0.00381 0.00443 -0.00121 -0.00446 -0.000321
(-0.26) (-0.62) (0.69) (-0.20) (-1.00) (-0.07)
separation 0.00353 -0.0109** 0.00349 -0.0121*** -0.00289 -0.00422
(0.72) (-2.35) (0.71) (-2.62) (-0.86) (-1.25)
_cons -0.119 -0.00179 -0.562 -0.193 0.205 0.0737
(-0.35) (-0.01) (-1.55) (-0.61) (0.92) (0.32)
Year &industry controlled controlled controlled controlled controlled controlled
Obs. 1401 1402 1401 1402 2803 2803
R-squared 0.1139 0.1188 0.1054 0.1262 0.1057 0.0958
Adj. R-squared 0.1023 0.1054 0.0987 0.1124 0.0947 0.0897

t statistics are in parentheses. * p < 0.10, ** p < 0.05, *** p < 0.01.
 DOES INDUSTRIAL POLICY PLAY AN IMPORTANT ROLE IN ENTERPRISE INNOVATION? 13

allocation is low, the industrial subsidy increases by 1% and the patent output increases by 0.039%,
and when the labor allocation is high under the support of industrial policy, the coefficient of
ip×subsidy_1 decreases from 0.0390 to 0.0356. For every 1% increase in industrial subsidies, patent
output increases by 0.0356% and is not significant. This shows that compared to high labor allocation
firms, firms with low labor allocation that are supported by industrial policies receive industrial
subsidies, which play a large role in improving innovation efficiency. This situation also occurs in the
variation of the coefficients of ip×tax-incentives_1 (see columns (3) and (4) of Table 4). That is,
compared to enterprises that are not supported by industrial policies, when labor allocation increases
from low to high, the coefficient of ip×tax-incentives_1 decreases from large (0.0853) to small
(−0.185) . This means that when tax incentives increase by 1%, innovation efficiency increases by
0.0853% with low labor allocation (L-labor) and decreases by 0.185% with high labor allocation
(H-labor), compared to enterprises not supported by industrial policies.
To obtain more detailed statistical evidence, we test the effectiveness of the benefit guidance
mechanism when we add an interaction item (subsidy_1 × labor_1，tax-incentives_1 × labor_1) to the
patent output equation in the sample of enterprises supported by industrial policy (ip = 1). This is
shown in Table 5.
Table 5 is a group regression based on different levels of labor allocation in the sample of firms supported
by industrial policies (ip = 1), which is used to test the relationship between industrial policies and enterprise
innovation in different groups of labor allocation. we divide the groups following Aghion, Hleimous, and
Kharroubi (2014), when the value of each characteristic index(labor) is higher than the median value group, it
is in the high value group (H-labor); otherwise, it is in the low value group (L-labor). In Table 5, the
dependent variable is lninvent, the core independent variables are subsidy_1 × labor_1, (subsidy_1 × labor_1)2,
tax-incentives_1 × labor_1, and (tax-incentives_1 × labor_1)2 and the other control variables are defined in
Table 1.
In Table 5, we show the moderating effect on the relationship between financial support and enterprise
innovation efficiency after labor allocation factors are added. Columns (1a) and (1b) in Table 5 show that
with the increase in labor allocation, the subsidy_1 × labor_1 coefficient decreases from large (0.00101) to
small (0.000227), and from significant to insignificant. That is, when labor allocation is low and
maintained, the industrial subsidy increases by 1% and the patent output increases by 0.00101%. When
labor allocation is high and maintains a certain level, the subsidy_1 × labor_1 coefficient increases by
0.00101%. For every 1% increase in industrial subsidies, patent output decreases by 0.000227%. This
situation also occurs in the coefficient of the interaction term tax-incentives_1 × labor_1, as shown in
columns (2a) and (2b) in Table 5.
To obtain further statistical evidence, we add the quadratic terms of the interaction term (subsidy _1
× labor_1)2 and (tax-incentives_1 × labor_1)2 to the innovation efficiency equation to test the inverted
U-shaped curve. As shown in column (3a) and (3b) in Table 5, the coefficients of (subsidy _1 × labor_1)2 and
(tax-incentives_1× labor_1)2 are significantly negative at the level of 0.01, that is, the moderating role of labor
allocation in the relationship between industrial policy and enterprise innovation efficiency is as follows that
the trend in changes first rises and then decreases, which further confirms H2.
To further reveal the regulatory mechanism of labor allocation in the relationship between industrial
policy and enterprise innovation efficiency, we test H3. The impact mechanism tests are in Table 6.
Table 6 examines the role of labor allocation in industrial policy and enterprise innovation in the sample
of firms supported by industrial policies (ip = 1). we divide the groups following Aghion, Hleimous, and
Kharroubi (2014), when the value of each characteristic index(labor, c_labor) is higher than the median
value group, it is in the high value group (H-labor, H-c_labor); otherwise, it is in the low value group
(L-labor, L-c_labor). Here, c_labor is the labor price measured as the employee’s annual average monthly
salary. In Table 6, the dependent variables are R&D, Labor, and lninvent, the core independent variables are
subsidy_1, tax-incentives_1, subsidy_1× lnRD_1, tax-incentives_1× lnRD_1, and c_labor_1 and the other
control variables are defined in Table 1.
Table 6. Mechanisms of the role of labor allocation in the relationship between industrial policy and patent output.
14

R&D lninvent

R&D labor H-c_labor L- c_labor H-c_labor L- c_labor H-c_labor L- c_labor

F. FENG

(1) (2) (3) (4) (5) (6) (7) (8) (9)

subsidy _1 0.893*** 0.104*** 0.0325

(24.41) (5.91) (0.64)
tax-incentives_1 0.302*** 0.168*** 0.0828
(3.72) (5.86) (1.11)
lnRD_1 0.470*** 0.385*** 0.487*** 0.398***
(11.98) (10.13) (12.75) (10.90)
subsidy _1×lnRD_1 0.0578*** 0.00626
(4.04) (0.41)
tax-incentives_1×lnRD_1 0.0737*** -0.00267
(3.81) (-0.12)
c_labor_1 -0.000216***
(-3.13)
labor_1 -0.00516* -0.00389
(-1.77) (-1.50)
leverage -0.0500*** -0.0629*** 0.420*** -0.0228** -0.0223** -0.000917 -0.000884 0.00307 0.000577
(-11.53) (-13.14) (4.99) (-2.05) (-2.09) (-0.49) (-0.49) (1.51) (0.29)
market-share -1.730*** -3.091*** 71.56*** -2.384 -5.658*** -0.835*** -0.619** -0.861*** -0.718**
(-2.81) (-4.58) (5.95) (-1.23) (-2.85) (-3.47) (-2.06) (-3.59) (-2.40)
state- share -0.00258 0.000215 0.0718 -0.0307*** -0.00441 0.000475 -0.00165 0.00102 -0.00123
(-0.40) (0.03) (0.57) (-3.40) (-0.48) (0.19) (-0.57) (0.40) (-0.42)
size -0.154 -0.140 -30.90*** -0.519 0.653** 0.0549 0.0448 0.00677 0.0407
(-1.37) (-1.14) (-13.95) (-1.41) (1.97) (1.02) (0.82) (0.12) (0.75)
diversification -0.0974 -0.0799 -2.521 0.165 0.709*** -0.0414 -0.172*** -0.0265 -0.164***
(-0.66) (-0.49) (-0.87) (0.55) (2.92) (-0.63) (-2.90) (-0.40) (-2.75)
ability-to-repay -0.0610*** -0.0624*** 0.745*** 0.00864 0.0346*** 0.000715 0.000301 0.00251 0.00150
(-9.84) (-9.13) (6.12) (0.65) (2.75) (0.26) (0.12) (0.90) (0.59)
e-concentration -0.00920 -0.0124 -0.154 -0.0177 -0.0242* -0.000323 -0.00680* -0.000854 -0.00703*
(-1.06) (-1.30) (-0.90) (-1.14) (-1.69) (-0.09) (-1.86) (-0.23) (-1.92)
control-competition 0.0132 0.0110 -0.334* -0.0471** 0.00840 0.00347 -0.00557 0.00196 -0.00684*
(1.47) (1.12) (-1.90) (-2.19) (0.44) (0.88) (-1.54) (0.49) (-1.87)
independence 0.0395*** 0.0661*** 0.617** 0.0205 -0.0290 -0.00415 -0.00493 -0.000376 -0.00245
(2.82) (4.30) (2.25) (0.82) (-1.32) (-0.68) (-0.86) (-0.06) (-0.43)
separation -0.00189 -0.0179 0.700*** -0.00434 0.0500 -0.00328 -0.00603 -0.00428 -0.00611
(-0.18) (-1.54) (3.37) (-0.09) (1.50) (-0.70) (-1.41) (-0.92) (-1.43)
_cons 5.562*** 6.703*** 137.8*** 8.815*** 4.274*** -0.281 0.289 -0.541* 0.123
(8.19) (8.98) (10.36) (5.22) (2.97) (-0.90) (1.00) (-1.68) (0.42)
Year &industry Controlled Controlled Controlled Controlled Controlled Controlled Controlled Controlled Controlled
Obs. 2803 2803 2803 1402 1401 1401 1402 1401 1402
R-squared 0.3486 0.2113 0.0925 0.1069 0.0499 0.2504 0.1587 0.2541 0.1557
Adj. R-squared 0.3129 0.1985 0.0877 0.0964 0.0412 0.2318 0.1324 0.2215 0.1368

t statistics are in parentheses. * p < 0.10, ** p < 0.05, *** p < 0.01.
DOES INDUSTRIAL POLICY PLAY AN IMPORTANT ROLE IN ENTERPRISE INNOVATION?
15
16 F. FENG

The relationship between financial support and R&D investment is shown in column (1) and (2) in
Table 6. R&D investment (RD) is expressed as the ratio of R&D expenditure to sales revenue. The
coefficients of subsidy_1 and tax-incentives_1 are significantly positive (0.893, 0.302). That is, R&D
investment increases by 0.893% and 0.302% respectively for every 1% increase in industrial
subsidies and tax preferences. This shows that industrial policy guidance promotes the increase in
R&D investment at enterprises that are supported by industrial policy. At the same time, the
relationship between labor cost (c_labor) and labor allocation (c_labor) is shown in Table 6 column
(3). The coefficient of the independent variable (c_labor_1) is negative, that is, when the labor cost
(c_labor_1) rises, the labor allocation decreases. In Table 6 column (4), shows that when the labor
cost is high (H-c_labor), the labor allocation is negatively correlated with R&D investment. The
results show that when the labor cost is high, firms reduce the use of labor and increase R&D
investment, thus forming an alternative relationship between labor investment and R&D investment.
However, when the labor cost is low (L-c_labor), the negative correlation between labor_1 and R&D
investment (RD) is not significant, as shown in column (5) in Table 6.
Table 6 columns (6) and (7) shows the relationship between R&D investment from fiscal stimulus
(industrial subsidies, tax incentives) and innovation efficiency at enterprises with different labor
costs. That is, when labor costs are high (H-c_labor), the coefficient of subsidy_1× lnRD_1 is
significantly positive, which indicates that when labor costs are high, the innovation efficiency of
R&D investment under fiscal incentives is enhanced, as shown in column (6) in Table 6. However,
when the labor cost is low (L-c_labor), the subsidy_1× lnRD_1 coefficient is not significant in column
(7) in Table 6. Similarly, the coefficients of tax-incentives_1× lnRD_1 have obvious differences when
the labor costs are higher (H-c_labor) and lower (L-c_labor): the former is significantly positive, the
latter is negative (not significant), and they have different coefficients, as shown in columns (8) and
(9) in Table 6.
Therefore, based on the statistical results in columns (1) to (9) in Table 6, we draw the following
conclusions. The channels through which labor allocation plays a role in the relationship between
industrial policy and enterprise innovation efficiency are: industrial policy is conducive to stimulating
enterprises’ incentives to increase R&D investment; enterprises with lower labor allocation tend to
have higher labor costs, but when labor cost is high, it will stimulate enterprises to increase R&D
investment, thus showing the substitution effect of R&D investment on labor investment. Under the
market pressure of higher labor cost, the interaction between industrial policy incentives and R&D
investment will further promote improvement in the innovation efficiency of enterprises. That is, the
statistical results confirm H3.

4.3. Robustness Test

To reduce the endogenous impact of the technical factors in industry’s innovation risk, in
a robustness test, we deleted the sample of high-tech enterprises and examined only non-high-
tech enterprises. In doing so, we follow Butler et al. (2011), Kostovetsky (2015), and Mao and
Xu (2016) based on the theory of industrial support to mitigate the insufficient incentives for
innovation activities and financing constraints. For the sample of non-high-tech enterprises, the
robustness test of the innovation output equation is carried out in the first decile, quintile, and
quartile interval samples of the return on assets (ROA) and credit financing (loan). Following
Acharya and Xu (2017), we select the number of invention patents obtained per unit of R&D
expenditure as an indicator of innovation output and use the variables above as a substitute
indicator of innovation efficiency. The dependent variable is tested for robustness. These
statistical results are in Tables 7 and 8.
Table 7 examines the impact of industrial interest stimulus on enterprise innovation efficiency
after deleting high-tech industry samples. The dependent variable is the natural logarithm of inven-
tion patents per unit of R&D investment (ln(invent/R&D)), the core independent variable is industrial
Table 7. Impact of industrial policy on enterprise innovation efficiency, based on regression results after deleting high-tech industry samples.
ln(invent/R&D)

Panel B Panel C

Sample by return on assets (ROA) Sample by credit financing (loan)

Panel A Q10 Q5 Q4 Q10 Q5 Q4

Non-high-tech industry full sample (1) (2) (3) (1) (2) (3)

subsidy _1 0.0575*** 0.0338 0.0480** 0.0541** 0.0312 0.0759*** 0.0903***

(3.51) (0.94) (2.03) (2.48) (0.70) (2.68) (3.11)
ROA_1 0.0198*** 0.00298 0.00761 0.00642 0.00738 0.00781 0.00571
(3.64) (0.21) (0.72) (0.65) (0.67) (1.00) (0.63)
loan_1 -0.00436* -0.00339 -0.00145 -0.00405 0.0329 0.0313*** 0.0263**
(-1.65) (-0.45) (-0.33) (-1.05) (1.52) (2.89) (2.39)
leverage -0.00202 -0.000318 -0.00300 -0.00363 0.00157 -0.00480 -0.00198
(-1.18) (-0.08) (-1.17) (-1.58) (0.34) (-1.39) (-0.62)
market-share -0.00541 -0.102 0.114 0.291 0.846 0.142 0.130
(-0.03) (-0.23) (0.47) (1.34) (1.58) (0.34) (0.39)
state- share -0.00343 0.0102* 0.00311 0.00152 0.00115 -0.000700 -0.00346
(-1.52) (1.71) (0.83) (0.46) (0.14) (-0.15) (-0.77)
size 0.296*** 0.324*** 0.300*** 0.284*** 0.130 0.270*** 0.329***
(7.61) (3.06) (5.11) (5.54) (1.08) (3.04) (4.38)
diversification -0.305*** -0.406*** -0.343*** -0.357*** -0.425** -0.188 -0.400***
(-5.74) (-2.70) (-3.93) (-4.65) (-2.50) (-1.58) (-3.73)
ability-to-repay -0.00108 -0.00621 -0.00119 -0.00302 0.00454 0.00409 0.00176
(-0.56) (-1.12) (-0.40) (-1.15) (0.80) (0.98) (0.47)
e-concentration -0.00668*** -0.00243 -0.00335 -0.00210 -0.00163 0.0000911 -0.00755*
(-2.74) (-0.41) (-0.94) (-0.65) (-0.22) (0.02) (-1.71)
control-competition -0.00285 -0.000265 0.00253 0.00111 0.00455 0.00791 0.000151
(-0.87) (-0.03) (0.45) (0.23) (0.43) (1.08) (0.02)
independence 0.00558 0.00644 0.0119 0.00897 -0.000682 0.0154 0.00292
(1.05) (0.38) (1.37) (1.16) (-0.04) (1.27) (0.27)
separation -0.000414 0.0199** 0.00649 0.00561 -0.0121 -0.00241 -0.000106
DOES INDUSTRIAL POLICY PLAY AN IMPORTANT ROLE IN ENTERPRISE INNOVATION?

(-0.12) (2.04) (1.10) (1.08) (-1.11) (-0.30) (-0.02)

17

(Continued )
 18
F. FENG

Table 7. Impact of industrial policy on enterprise innovation efficiency, based on regression results after deleting high-tech industry
samples. (Continued)
ln(invent/R&D)

Panel B Panel C

Sample by return on assets (ROA) Sample by credit financing (loan)

Panel A Q10 Q5 Q4 Q10 Q5 Q4

Non-high-tech industry full sample (1) (2) (3) (1) (2) (3)

_cons 0.284 -0.0951 -0.132 0.117 0.881 -0.136 0.351

(1.13) (-0.13) (-0.33) (0.33) (1.03) (-0.24) (0.69)
Year &industry Controlled Controlled Controlled Controlled Controlled Controlled Controlled
Obs. 2564 256 512 641 256 512 641
R-squared 0.0732 0.1575 0.0885 0.0869 0.1095 0.0877 0.1054
Adj. R-squared 0.0612 0.1433 0.0801 0.7955 0.0981 0.0802 0.0913

t statistics are in parentheses. * p < 0.10, ** p < 0.05, *** p < 0.01.
Table 8. Impact of industrial policy on enterprise innovation efficiency: based on regression results after deleting high-tech industry samples.
ln(invent/R&D)

Panel B Panel C

Sample by return on assets (ROA) Sample by credit financing (loan)

Panel A Q10 Q5 Q4 Q10 Q5 Q4

Non-high-tech industry full sample (1) (2) (3) (1) (2) (3)

tax-incentives_1 0.367*** -0.183 0.0128 0.274*** 0.296*** 0.309*** 0.306***

(10.67) (-1.55) (0.11) (3.30) (2.86) (4.03) (4.37)
ROA_1 -0.00646 0.0157 0.0382** -0.0161 -0.00698 -0.0122 -0.0129
loan_1 (-1.11) (1.01) (2.56) (-1.34) (-0.58) (-1.17) (-1.31)
leverage -0.00506* -0.00407 -0.00123 -0.00383 0.0264 0.0295** 0.0236**
(-1.95) (-0.55) (-0.25) (-1.00) (1.24) (2.29) (2.15)
market-share 0.000762 -0.000206 -0.0165** -0.00358 0.00218 0.00139 0.000378
(0.45) (-0.05) (-2.35) (-1.56) (0.48) (0.41) (0.12)
state- share 0.267* 0.0695 0.187 0.329 0.942* 0.604 0.253
(1.68) (0.15) (0.18) (1.52) (1.79) (1.58) (0.75)
size -0.00380* 0.00944 -0.000320 0.00108 0.00201 0.000338 -0.00295
(-1.72) (1.59) (-0.05) (0.33) (0.25) (0.07) (-0.66)
diversification 0.281*** 0.297*** -0.146 0.290*** 0.127 0.226*** 0.305***
(7.36) (2.78) (-0.34) (5.68) (1.08) (2.74) (4.07)
ability-to-repay -0.316*** -0.430*** 0.0965 -0.365*** -0.411** -0.388*** -0.423***
(-6.07) (-2.87) (0.60) (-4.76) (-2.46) (-3.25) (-3.98)
e-concentration 0.00105 -0.00530 0.0160* -0.000996 0.00526 0.00466 0.00498
(0.56) (-0.96) (1.68) (-0.38) (0.95) (1.16) (1.35)
control-competition -0.00580** -0.00216 -0.000884 -0.00184 -0.000204 -0.00562 -0.00635
(-2.43) (-0.37) (-0.14) (-0.57) (-0.03) (-1.13) (-1.45)
-0.00489 -0.00178 -0.0175 0.000741 0.00279 0.000458 -0.00173
(-1.53) (-0.19) (-1.31) (0.15) (0.27) (0.06) (-0.26)
independence 0.00374 0.00761 -0.0148 0.00812 -0.00731 -0.00104 -0.000453
(0.72) (0.45) (-0.92) (1.06) (-0.44) (-0.08) (-0.04)
separation -0.000753 0.0190* -0.0409* 0.00429 -0.0118 -0.00438 0.000107
DOES INDUSTRIAL POLICY PLAY AN IMPORTANT ROLE IN ENTERPRISE INNOVATION?

(-0.22) (1.95) (-1.77) (0.83) (-1.11) (-0.57) (0.02)

19

(Continued )
 20
F. FENG

Table 8. Impact of industrial policy on enterprise innovation efficiency: based on regression results after deleting high-tech industry
samples. (Continued)
ln(invent/R&D)
Panel B Panel C

Sample by return on assets (ROA) Sample by credit financing (loan)

Panel A Q10 Q5 Q4 Q10 Q5 Q4

Non-high-tech industry full sample (1) (2) (3) (1) (2) (3)

_cons 0.187 -0.0265 3.228* 0.117 0.948 0.589 0.367

(0.76) (-0.04) (1.88) (0.33) (1.13) (1.03) (0.73)
Year &industry Controlled Controlled Controlled Controlled Controlled Controlled Controlled
Obs. 2564 256 512 641 256 512 641
R-squared 0.1137 0.1637 0.1185 0.0950 0.1402 0.1230 0.1200
Adj. R-squared 0.1011 0.1589 0.0958 0.0892 0.1387 0.1154 0.1131

t statistics are in parentheses. * p < 0.10, ** p < 0.05, *** p < 0.01.
 DOES INDUSTRIAL POLICY PLAY AN IMPORTANT ROLE IN ENTERPRISE INNOVATION? 21

subsidies (subsidy _1), and the other control variables are defined in Table 1. The return on assets
(ROA) is the net profit at the end of the period/total assets at the end of the period, and credit
financing (loan) is the incremental amount of loans/total assets. The increase in loans here is the sum
of the increase in short-term loans and in long-term loans. Q10, Q5, and Q4 are the first decile,
quintile, and quartile samples, respectively, based on the classification index.
Table 8 examines the impact of industrial interest stimulus on enterprise innovation efficiency
after deleting high-tech industry samples. The dependent variable is the natural logarithm of inven-
tion patents per unit of R&D investment (ln(invent/R&D)), the core independent variable is industrial
subsidies (tax-incentives_1), and the other control variables are defined in Table 1.The return on
assets (ROA) is the net profit at the end of the period/total assets at the end of the period, and credit
financing (loan) is the incremental amount of loans/total assets. The increase in loans here is the sum
of the increase in short-term loans and in long-term loans. Q10, Q5, and Q4 are the first decile,
quintile, and quartile samples, respectively, based on the classification index.
In Table 7, the coefficient of subsidy_1 in panel A is significantly positive in the full sample, and in
the subsample, with Q10, Q5, and Q4. The significance of the coefficient of the independent variable
(subsidy _1) is also increasing, from not significant to significant, indicating under the stimulus of
fiscal interests, the innovation efficiency of enterprises has been significantly improved (as shown in
panel B of Table 7). These results are also presented in Table 7, panel C. In the full sample (panel A)
of Table 8, the coefficient of tax-incentives_1 is also significantly positive, which shows the positive
effect of fiscal incentives on the efficiency of enterprise innovation. Similarly, in the subsample, as
the sample size of Q10, Q5, and Q4 gradually expands, the significance of the coefficient of the
independent variable (tax-incentives_1) also increases from insignificant to significant, and the
coefficient changes from small to large and negative to positive. This further demonstrates that,
under the stimulus of financial interests, the innovation efficiency of enterprises has significantly
improved (as shown in panel B of Table 8). In panel C of Table 8, in the samples by credit financing
(loan), the coefficients of the independent variables (tax-incentives_1) are all significantly positive.
Therefore, Tables 7 and 8 show that the positive impact of industrial subsidies (subsidy _1) and
income tax incentives (tax-incentives_1) on the innovation efficiency of enterprises has not changed
fundamentally. This robustness test confirms the robustness of our earlier conclusions obtained
through grouping and regression analysis.

5. Main Conclusions and Policy Implications

Using a sample of A-share listed companies in Shanghai and Shenzhen during the twelfth FYP period
(2011–2015), we study the impact of industrial policy and labor allocation on enterprise innovation
efficiency, to test the link between government policy, labor allocation, and innovation efficiency.
Our empirical results show that industrial policy significantly improves the innovation efficiency of
enterprises. At enterprises with low labor allocation, the innovation efficiency of enterprises is higher
with the support of industrial policy; and the mechanism of the adjustment role of labor allocation in
industrial policy and enterprise innovation is R&D investment caused by the labor cost effect. The
substitution effect of capital on the labor force is that industrial policy motivates enterprises to
increase R&D investment. Enterprises with a lower labor force allocation tend to have higher labor
cost, and when the labor cost is high, it stimulates enterprises to increase R&D investment, thus
showing the substitution effect of R&D investment on labor investment and at higher labor cost.
Under market pressure, the interaction between industrial policy incentives and R&D investment has
greatly promoted improvement in the innovation efficiency of enterprises. This study provides a new
perspective and empirical evidence for government decision-makers to understand the relationship
between government policies, labor market factors, and enterprise innovation.
These conclusions indicate that industrial policy improves the efficiency of enterprise innovation through
fiscal incentives, and factors such as labor allocation, which are closely related to the labor market, have an
22 F. FENG

important impact. The important mechanism is that financial support stimulates enterprises to increase R&D
expenditure. However, higher labor costs encourage enterprises to increase R&D investment and innovation
to replace labor. That is, when labor costs are high, the effect of labor cost makes the industrial policy
support play a positive role in the innovation efficiency of enterprises. Therefore, the guidance of industrial
policy through fiscal incentives helps promote enterprise R&D investment, and when labor is scarce, it can
enhance the effect of industrial policy on enterprise innovation efficiency. The government is thus advised to
tap the positive effect of industrial policies and measures and guide enterprises with higher labor costs to
carry out technological innovation, which has great and far-reaching practical significance for driving
technological transformation and upgrading enterprises.

Funding
This paper is supported by the following project programs. Foundation projects: National Natural
Science Foundation projects (71272196, 71572206); Guilin Tourism University High-level Talents
Research Initiation Fund projects (KJ0613004), Guilin Tourism University major topics (2018MP01),
and “first-class discipline construction” projects (CZ6101002).

Notes
1. Here we mainly refers to the non-monopoly of knowledge innovation income.
2. Administrative examination, also known as administrative confirmation, is the process in which admin-
istrative agencies examine and approve the legitimacy and authenticity of the administrative acts for example, to
set up a company, it is necessary to submit the identity card of the owner of the company, the valid financial
certificate of the registered funds, and the application materials for the scope of business, and to verify the
certification; administrative approval, also known as administrative license, means that the administrative subject
agrees that a specific counterpart obtains certain legal qualifications or implements certain acts, in practice this is
the issuance of licenses, for example, in China, natural resources such as land and mines are owned by the state.
If a private company wants to obtain the right to use or exploit, it must obtain the consent, permission and
corresponding business license from the competent government departments.
3. Enterprise autonomy refers to the right of state-owned enterprises to operate independently. Recognition
and protection of enterprise autonomy ensures not only the unity of the national economy because China’s state-
owned enterprises control the country’s main economic lifeline, such as mines, land and other natural resources,
communications, roads, civil aviation, finance, electricity, water, oil, natural gas and other livelihood industries,
are all controlled by the government’s state-owned enterprises. The operation of state-owned enterprises is often
guided and controlled by government policies to ensure the integrity of the national economy but also the
diversity and flexibility of enterprise production and operations. Under China’s traditional economic manage-
ment system, enterprises lack the necessary autonomy. After the Third Plenum Session of the Eleventh Central
Committee of the Chinese Communist Party in 1978, the State Council issued enterprise reforms, which clearly
stipulated the responsibilities, rights, and interests of enterprises and expanded their autonomy.
4. “Innovation numbness” refers to disinterest in innovation while “joint innovation” indicates following of
leading innovators in carrying out technological change.
5. After innovative knowledge is made public, few people pay for it. This is called the non-exclusivity of revenue.
6. ST is abbreviated as Special Treatment, meaning “Special treatment” .ST company is a listed company
with financial or other abnormal conditions.PT is the abbreviation of Particular Transfer. According to the
“company law” and “securities law”, if a listed company suffers losses for three consecutive years, its shares will
be suspended from listing. Since July 9, 1999, the Shanghai and Shenzhen stock exchanges have implemented
a special transfer service for such suspended stocks, which is prefixed with PT and called PT stock. If a listed
company loses money for 2 consecutive years, loses money for 1 consecutive year, and its net assets break the
buck, or the company has a major illegal act in the process of operation, the stock exchange will carry out special
treatment on the company’s stock, namely the ST system. For ST company, if problems occur again, such as
continuing losses in the next year to reach the 3 consecutive years loss limit in the company law, PT treatment
will be carried out.
7. Whether the industry receives government support depends on its attitude toward development of that
particular industry. For instance, the Chinese government has adopted policies to restrict the development of
highly-polluting mining and chemical industries, while encouraging high-tech industries, such as integrated
circuits and new energy.
 DOES INDUSTRIAL POLICY PLAY AN IMPORTANT ROLE IN ENTERPRISE INNOVATION? 23

References
Acemoglu, D. 2002. Directed technical change. Review of Economic Studies 69:781–809. doi:10.1111/roes.2002.69.issue-4.
Acemoglu, D. 2010. When does labor scarcity encourage innovation? Journal of Political Economy 118:1037–78.
Acharya, V., and Z. X. Xu. 2017. Financial dependence and innovation: The case of public versus private firms. Journal of
Financial Economics 124:223–43. doi:10.1016/j.jfineco.2016.02.010.
Aghion, P., D. Hleimous, and E. Kharroubi. 2014. Cyclical fiscal policy, credit constraints, and industry growth. Journal of
Monetary Economics 62:41–58. doi:10.1016/j.jmoneco.2013.12.003.
Aghion, P., M. Dewatripont, L. Du, A. E. Harrison,and, and P. Legros. 2012. Industrial policy and competition. CEPR
Discussion Papers no.7. doi:10.1094/PDIS-11-11-0999-PDN.
Bena, J., and E. Simintzi. 2017. Globalization of work and innovation: Evidence from doing business in China. Discussion
paper no.319. Harvard business school.
Butler, A. W., J. Cornaggia, G. Grullon, and J. P. Weston. 2011. Corporate financing decisions, managerial market timing, and
real investment. Journal of Financial Economics 101:666–83. doi:10.1016/j.jfineco.2011.05.001.
Chen, D., Z. Li, and F. Xin. 2013. Five-year plans of China in finance and their consequences. Nanjing University Working
Paper.
Habakkuk, H. J. 1962. American and British Technology in the Nineteenth Century: The search for labour saving inventions.
Cambridge: Cambridge University Press.
Ham, R. M., and D. C. Mowery. 1998. Improving the effectiveness of public-private R&D collaboration: Case studies at a US
weapons laboratory. Research Policy 26:661–75. doi:10.1016/S0048-7333(97)00041-3.
Hanlon, W. W. 2015. Necessity is the mother of invention: Input supplies and directed and directed Technical change.
Econometrica 83:67–100. doi:10.3982/ECTA10811.
Hicks, J. R. 1932. The theory of wages. London: Macmillan.
Hornbeck, R., and S. Naidu. 2014. When the levee breaks: Black migration and Economic development in the American south.
American Economic Review 104:963–90. doi:10.1257/aer.104.3.963.
Kerr, W. R., and R. Nanda. 2015. Financing innovation. Annual Review of Financial Economics 7:445–62. doi:10.1146/
annurev-financial-111914-041825.
Kostovetsky, L. 2015. Political capital and moral hazard. Journal of Financial Economics 116:144–59. doi:10.1016/j.
jfineco.2014.12.003.
Lazzarini, S. G. 2015. Strategizing by the government: Can industrial policy create firm-level competitive advantage? Strategic
Management Journal 17:97–112. doi:10.1002/smj.2204.
Li, W. J., and M. N. Zheng. 2016. Substantive innovation or strategic innovation? The impact of macro-industrial policy on
micro-enterprise innovation. Economic Research 4:60–73. (in Chinese).
Lu, Z. F., and F. C. Han. 2013. How does macroeconomic policy affect the economic effect of cash holding? based on the study
of product market and capital market. The World of Management 6:43–60. (in Chinese).
Mao, Q. L., and J. Y. Xu. 2016. Government subsidies, heterogeneity,and enterprise risk-taking. Economics (Quarterly)
4:1533–62. (in Chinese).
Mao, Y. F., and J. J. X. Wang. 2018. Labor scarcity, finance, and innovation: Evidence from antebellum America. Kelley
School of Business Research Paper No. 18.
Mukherjee, A., M. Singh, and A. Aldokas. 2017. Do corporate taxes hinder innovation? Journal of Financial Economics 124
(1):195–221. doi:10.1016/j.jfineco.2017.01.004.
Peneder, M. 2008. The problem of private under-investment in innovation: A policy mind map. Technovation 28:518–30.
doi:10.1016/j.technovation.2008.02.006.
Rosenberg, N. 1969. The direction of technological change: Inducement mechanisms and focusing devices. Economic
Development and Cultural Change 18:1–24. doi:10.1086/450399.
Song, L. Y., and X. B. Wang. 2013. Key industrial policies, resource replacement,and industrial productivity. The World of
Management 12:63–77. (in Chinese).
Spence, M. 1984. Cost reduction, competition, and industry performance. Econometrica 52:101–21. doi:10.2307/1911463.
Tuzel, S., and M. B. Zhang. 2018. Economic stimulus at the expense of routine-task Jobs. University of Southern California,
Working Paper.
Vorley, T., and J. Nelles. 2010. Innovation policy as industrial policy: Some lessons from Hamburg’s regional innovation
system. Local Economy: the Journal of the Local Economy Policy Unit 25 (8):631–49. doi:10.1080/02690942.2010.533422.
Wrigley, L. 1970. Division autonomy, and diversification. PhD diss, Harvard University.
Zhou, Y. H., S. Puyulu, Y. Chen, and F. Fang. 2015. Government support and new industry development: Take new energy as
an example. Economic Research 6:147–61. (in Chinese).